Missed TensorFlow World? Check out the recap. Learn more

tf.nn.l2_loss

TensorFlow 2.0 version

Defined in generated file: python/ops/gen_nn_ops.py

L2 Loss.

Aliases:

  • tf.compat.v1.nn.l2_loss
  • tf.compat.v2.nn.l2_loss
tf.nn.l2_loss(
    t,
    name=None
)

Computes half the L2 norm of a tensor without the sqrt:

output = sum(t ** 2) / 2

Args:

  • t: A Tensor. Must be one of the following types: half, bfloat16, float32, float64. Typically 2-D, but may have any dimensions.
  • name: A name for the operation (optional).

Returns:

A Tensor. Has the same type as t.